Is glGetError much faster lately than rumor has it

I notice generic advice is "don’t use lots of GetError!).

But I noticed on a NVIDIA 8600M GT on latest drivers I get exactly or almost exactly the same FPS when using lots of them (including one on each glDraw… ) or none at all. I even made a pointless loop of 100s of it and no change.

Though the application does not utilize all of the resources of the hardware, e.g. no textures yet and I suppose no much VRAM yet hence; could it be that?

[Though I’m skeptical it’s merely no full utilization of VRAM since the FPS does go down if e.g. vertex shader is loaded with more calculations].

Is there some kind of ‘special section of the hardware’ that would be overloaded by that operation and it’s usually low on load when GetError is used unless other effects load it?

speaking about AMD implementation:
the main issue with glGetError() is that it prevents efficient multithreading inside the opengl driver, however it does not impact HW performance.

But I noticed on a NVIDIA 8600M GT on latest drivers I get exactly or almost exactly the same FPS when using lots of them (including one on each glDraw… )

Are you rendering anything that’s CPU-bound? That is, is your code stressful enough that calling glGetError would be a problem?

Nvidia drivers have / used to have a switch that would silently turn off OpenGL error reporting. Check your Nvidia Control Panel for something like that.